DynamoDB Capacity Mode Cost Thresholds for Low-Traffic Workloads

Answer the exact question first

The break‑even point is the monthly request volume where the cost of running a Provisioned table (including Auto‑Scaling “minimum” capacity) equals the cost of the same traffic in On‑Demand mode.

1. Break‑Even for Standard (Hot) Table Class

Assumptions (US East (N. Virginia) – values are region‑specific):

• On‑Demand read request price: $1.25 per 1 000 000 reads
• On‑Demand write request price: $1.25 per 1 000 000 writes
• Provisioned read capacity unit (RCU) price: $0.00065 per hour
• Provisioned write capacity unit (WCU) price: $0.00130 per hour
• 1 RCU ≈ 4 strongly consistent reads per second (or 8 eventually consistent reads)
• 1 WCU ≈ 1 write per second

For a workload that is read‑heavy and has a steady average of R reads per second, the minimum provisioned RCU needed is ceil(R / 4). The monthly cost (≈30 days) is:

RCU_cost = ceil(R / 4) × 24 × 30 × 0.00065
OnDemand_cost = R × 86_400 × 1.25e-6

Setting the two equal and solving for R gives a break‑even of roughly 0.005 reads/s (≈17 reads/minute). Below that, On‑Demand is cheaper; above it, Provisioned with Auto‑Scaling becomes more economical.

For a write‑heavy workload with average W writes per second, the threshold is about 0.002 writes/s (≈7 writes/minute).

2. Effect of Standard‑IA Table Class

Standard‑IA lowers storage costs but increases the per‑request price. The request price is roughly doubled compared to Standard:

• On‑Demand read: $2.50 per 1 000 000 reads
• On‑Demand write: $2.50 per 1 000 000 writes
• Provisioned RCU/WCU prices stay the same

Re‑solving the equations with the new request price shifts the break‑even upward. For reads, the threshold climbs to about 0.01 reads/s (≈30 reads/minute). For writes, it rises to roughly 0.004 writes/s (≈12 writes/minute). In other words, the higher request cost pushes the break‑even point to a higher traffic level, making Provisioned mode attractive only when traffic is somewhat higher than in the Standard class.

3. Practical Verification Steps

1. Use the AWS Pricing Calculator to input your exact region’s RCU/WCU and request prices.
2. Enter your expected average reads/writes per second and the read/write consistency mix.
3. Compare the projected monthly costs for both Provisioned (with a minimum capacity that matches the average) and On‑Demand.
4. For tables that may experience bursts, add a safety margin (e.g., 20‑30 % above the average) to the Provisioned minimum.
5. Monitor CloudWatch metrics (ConsumedReadCapacityUnits and ConsumedWriteCapacityUnits) over a month to confirm that the chosen Provisioned capacity stays above the actual peak.

4. Missing diagnostic detail that could change the recommendation

Could you share the read/write ratio and whether the traffic is strongly consistent or eventually consistent? Those details influence the required RCUs and therefore the break‑even calculation.